4 files changed, 17 insertions, 0 deletions
diff --git a/extractor/rule_factory.cc b/extractor/rule_factory.cc
index 4101fcfa..fbdc7cce 100644
--- a/extractor/rule_factory.cc
+++ b/extractor/rule_factory.cc
@@ -220,6 +220,8 @@ Grammar HieroCachingRuleFactory::GetGrammar(const vector<int>& word_ids) {
     }
   }
 
+  cerr << "Vocabulary size = " << vocabulary->Size() << endl;
+
   Clock::time_point stop_time = Clock::now();
   cerr << "Total time for rule lookup, extraction, and scoring = "
        << GetDuration(start_time, stop_time) << " seconds" << endl;
diff --git a/extractor/run_extractor.cc b/extractor/run_extractor.cc
index 38f10a5f..2b01e832 100644
--- a/extractor/run_extractor.cc
+++ b/extractor/run_extractor.cc
@@ -31,6 +31,14 @@ namespace fs = boost::filesystem;
 namespace po = boost::program_options;
 using namespace std;
 
+void my_pause() {
+  cerr << "pausing..." << endl;
+  for (int i = 0; i < 10000000; ++i) {
+    cerr << endl;
+  }
+  cerr << "end pause" << endl;
+}
+
 int main(int argc, char** argv) {
   // TODO(pauldb): Also take arguments from config file.
   po::options_description desc("Command line options");
diff --git a/extractor/vocabulary.cc b/extractor/vocabulary.cc
index 5c379a29..b68d76a9 100644
--- a/extractor/vocabulary.cc
+++ b/extractor/vocabulary.cc
@@ -24,3 +24,8 @@ bool Vocabulary::IsTerminal(int symbol) {
 string Vocabulary::GetTerminalValue(int symbol) {
   return words[symbol];
 }
+
+int Vocabulary::Size() {
+  return words.size();
+}
+
diff --git a/extractor/vocabulary.h b/extractor/vocabulary.h
index c6a8b3e8..ff3e7a63 100644
--- a/extractor/vocabulary.h
+++ b/extractor/vocabulary.h
@@ -19,6 +19,8 @@ class Vocabulary {
 
   virtual string GetTerminalValue(int symbol);
 
+  int Size();
+
  private:
   unordered_map<string, int> dictionary;
   vector<string> words;